A modular architecture for diagnosis of discrete event systems

Abstract

We introduce a modular architecture for diagnosis in the framework of discrete event systems. The architecture is best suited for systems with multiple interacting components. We suggest to use local diagnosis at each component and provide sufficient conditions so that the global system behavior can be diagnosed using these local diagnoses. Moreover, in the context of the presented modular architecture, we discuss an incremental algorithm to check diagnosability of systems in the framework of Sampath et al. (1995). The suggested algorithm has the advantage of being more computationally efficient than the one presented in the above paper.